How to manage feature flags without making performance worse

The Problem

Using libraries like Flagr to help maintain feature flags had a negative performance impact; it was taking 900ms to only check 50 flags.

Using Sentry performance monitoring, we discovered the majority of the time was spent deserializing the Flagr response.

The Solution

Digging into the client, we noticed that deserialization was turning the entire JSON blob (i.e. the flag results) into a Python object. While deserializing JSON is convenient from a development perspective, because we were only using a few JSON fields, the cost just wasn’t worth it.

After removing caching and patching our own deserializer with a simple json.loads(response.data) we saw a huge improvement in speed — so much so that other areas began to look slow by comparison:

Further Reading

• Using Sentry Performance To Make Sentry Performant
• Fetch Feature Flags Faster - Snack of the Week
• Feature Flag Docs